1. Home
  2. Insights
  3. Performance

Why a One-Second Delay Costs More Than You Think

Performance is a revenue lever. How we put a dollar figure on slowness and fix it in a single sprint.


Every team agrees the site should be faster. Few can say what slowness costs, so performance work loses to features in every planning meeting. Here is how we make the cost visible, and the fixes that pay back fastest.

Put a number on it

Take your analytics for the last 90 days and segment conversion rate by page load time bucket. In almost every dataset we have looked at, conversion falls as load time rises, and the drop between the fastest and slowest quartile is rarely under 20%. Multiply the gap by your traffic and average order value and you have a monthly figure that belongs in the business case.

Measure what users feel

Lab scores are useful for debugging but field data decides outcomes. Use the Chrome User Experience Report in Search Console, or your own real-user monitoring, and track three things:

  • Largest Contentful Paint: when the main content appears. Target under 2.5 seconds.
  • Interaction to Next Paint: how quickly the page responds to taps and clicks. Target under 200 milliseconds.
  • Cumulative Layout Shift: how much the page jumps around. Target under 0.1.

The fixes that pay back in one sprint

Images

Still the biggest win on most marketing sites. Serve modern formats, size images to their display width with srcset, lazy-load anything below the fold, and give every image explicit dimensions so the layout does not shift.

Third-party scripts

Audit every tag. Chat widgets, heatmaps, A/B tools and ad pixels routinely add a second or more of main-thread time. Remove what nobody looks at, defer the rest, and load chat on interaction rather than on page load.

Fonts

Self-host, subset to the characters you use, preload the one or two files needed above the fold and use font-display: swap. Two families and four weights is almost always enough.

JavaScript

Check the bundle for libraries that exist to do one small thing. Split routes, remove polyfills for browsers you no longer support, and render content on the server so the page is useful before scripts arrive.

Caching and the edge

Static assets should be immutable with a one-year cache. HTML should be served from an edge network with sensible revalidation. Both are configuration, not engineering.

Keep it fast

Performance regresses quietly. Add a Lighthouse check to your pull-request pipeline with a budget, and fail the build when a page exceeds it. The cost of the check is minutes; the cost of the regression is the number you worked out in the first section.